Molecular Biology of the CellVol. 6,1769-1780, December 1995

BET3 Encodes a Novel Hydrophilic Protein that Actsin Conjunction with Yeast SNAREsGuendalina Rossi, Karin Kolstad, Shelly Stone, Francois Palluault, andSusan Ferro-Novick*

Department of Cell Biology and Howard Hughes Medical Institute, Yale University School ofMedicine, New Haven, Connecticut 06510

Submitted August 2,1995; Accepted September 1, 1995Monitoring Editor: David Botstein

Here we report the identification of BET3, a new member of a group of interacting geneswhose products have been implicated in the targeting and fusion of endoplasmic retic-ulum (ER) to Golgi transport vesicles with their acceptor compartment. A temperature-sensitive mutant in bet3-1 was isolated in a synthetic lethal screen designed to identifynew genes whose products may interact with BETI, a type II integral membrane proteinthat is required for ER to Golgi transport. At 37°C, bet3-1 fails to transport invertase,a-factor, and carboxypeptidase Y from the ER to the Golgi complex. As a consequence,this mutant accumulates dilated ER and small vesicles. The SNARE complex, a docking/fusion complex, fails to form in this mutant. Furthermore, BET3 encodes an essential22-kDa hydrophilic protein that is conserved in evolution, which is not a component ofthis complex. These findings support the hypothesis that Bet3p may act before theassembly of the SNARE complex.

INTRODUCTION

Proteins are transported from the ER to the cell surfacethrough a series of membrane-bound compartments.In the case of endoplasmic reticulum (ER) to Golgitransport, specific carrier vesicles have been shown tomediate membrane traffic at this stage of the secretorypathway. Two soluble yeast gene products, SEC18 andSEC17, have been implicated in vesicle fusion (Wilsonet al., 1989; Griff et al., 1992). SEC18, which was origi-nally identified as a gene whose product is requiredfor vesicular transport between the ER and Golgi com-plex (Novick et al., 1980), also plays a role in intra-Golgi and post-Golgi membrane traffic events (Gra-ham and Emr, 1991). The mammalian homologue ofSEC18 is the N-ethylmaleimide-sensitive factor (NSF;Wilson et al., 1989), a cytosolic protein that mediatesthe fusion of vesicles with their acceptor compartment(Malhotra et al., 1988). The association of NSF withmembranes is dependent upon the presence of threesoluble NSF attachment proteins (a, ,B, and y SNAPs)

* Corresponding author: Howard Hughes Medical Institute, YaleUniversity School of Medicine, 295 Congress Avenue, BCMM254B, New Haven, CT 06510.

and a set of SNAP receptors that reside on the trans-port vesicle and target membrane (Clary et al., 1990).In yeast, SEC17 encodes a homologue of a-SNAP(Clary et al., 1990). The SNAP receptors, calledSNAREs, form a complex with NSF and the SNAPs (aand y). In neurons, the vesicle SNARE (v-SNARE) issynaptobrevin, while the SNAREs on the target mem-brane (t-SNARE) are syntaxin and SNAP-25. Each in-tracellular fusion event may have one set of v-SNAREsand t-SNAREs that interact with each other to ensurethe specificity of vesicular transport (Sollner et al.,1993; Warren, 1993; Ferro-Novick and Jahn, 1994).BETI, BOSi, SEC21, SEC22, and YPT1 are members

of a group of interacting yeast genes (Newman et al.,1990; Dascher et al., 1991). The SEC21 gene product isa component of yeast coatomer (Hosobuchi et al.,1992), while BETI, BOS1, and SEC22 encode cytoplas-mically oriented type II integral membrane proteinsthat are required for ER to Golgi transport (Newmanand Ferro-Novick, 1987; Dascher et al., 1991; Shim etal., 1991). Yptlp is a small GTP-binding protein thatalso functions at this stage of the secretory pathway(Ferro-Novick and Novick, 1993). Previous studieshave shown that Boslp, the v-SNARE of ER to Golgitransport vesicles, physically interacts with Sec22p on

© 1995 by The American Society for Cell Biology 1769

G. Rossi et al.

vesicles and not on other compartments. This interac-tion, which appears to facilitate the activity of Boslp,is catalyzed by Yptlp (Lian et al., 1994). Based on thesedata it was proposed that Yptlp, and other smallGTP-binding proteins, regulate the specificity of vesic-ular transport by selectively activating the v-SNAREon vesicles.In an effort to identify additional components that

may functionally interact with this group of proteins,we have used the betl-l mutant in a synthetic lethalscreen. Synthetic lethality results when the effect ofcombining two mutations in the same haploid straincauses cell death under growth conditions that arepermissive for either single mutant. Thus, the com-bined effect of both mutations impairs a specific pro-cess to a greater extent than either one alone. Here wereport the use of such a screen to identify a newtemperature-sensitive (ts) secretory mutant that isblocked in ER to Golgi transport, bet3-1. Like BETI,BET3 interacts genetically with BOSi, SEC21, SEC22,YPT1, and encodes an essential hydrophilic 22-kDaprotein that is highly homologous to a Caenorhabditiselegans protein (p20) of unknown function. Studiespresented here suggest that Bet3p functions before thepairing of the v-SNARE with its t-SNARE.

MATERIALS AND METHODS

Mutant Isolation and ScreenSFNY185 was grown to early exponential phase at 25°C to obtain 5x 107 cells/ml. Cells were washed twice with sterile water, resus-pended in 0.1 M sodium phosphate buffer (pH 7.0) and mu-tagenized with 3% ethylmethanesulfonate (-70% killing). Aftershaking for 1 h at 30°C, the cells were washed with sterile water,transferred to a new tube, and washed two more times with 5%sodium thiosulfate. Afterward, the cells were allowed to recoverduring a 20-h incubation at 25°C in 100 ml of YPD medium (totalOD599 of 25.5). Cells were centrifuged, resuspended in fresh YPDmedium, and then shifted to 37°C for 3 h. Subsequent to this shiftthe cells were washed, resuspended in 0.5 ml of water, and layeredonto a 12.5-ml Percoll gradient (90% Percoll prepared in yeastnitrogen base without amino acids) in a Beckman Quickseal centri-fuge tube (16 x 76 mm; Fullerton, CA). The tubes were heatsealedand spun at 22,000 x g for 20 min at 4°C in a Beckman ultracentri-fuge (TI50 rotor). Fractions (the densest 0.8-2.6% of the cells) werecollected (1.0 ml) using an ISCO fraction collector, diluted 2 x 10-4in sterile water, and 0.2 ml was plated onto selective plates that weregrown for 3 days at 25°C as described previously (Novick et al.,1980; Finger and Novick, 1992). The colonies were replica platedonto two different YPD plates and incubated at 25°C or 37'C for 1day. Temperature-sensitive mutants were isolated, purified, andcells that failed to grow after 5 days at 25°C on 5-FOA (5-fluoroo-rotic acid) plates were screened for defects in the secretion ofinvertase.

Antibody Production, Immunoprecipitation,and ElectrophoresisCells were grown overnight to an OD599 = 1.0 in yeast nitrogen basethat was supplemented with the appropriate amino acids and 2%glucose. Mutant and wild-type cells were resuspended in freshmedium to an OD599 = 3.0 and then preshifted at 37°C for 30 min.

At the end of this incubation, 200 ,uCi of I'S In Vitro label (Amer-sham, Arlington Heights, IL) was added to each sample and theincubation was continued for 30 min longer at 37°C. To examine thesecretion of invertase, cells were grown to an OD599 = 1.0 and thenpreshifted to 37°C for 30 min. Invertase synthesis was derepressedin minimal medium with 0.1% glucose at an OD599 = 2.0 during a30-min incubation at 37°C in the presence of 330 ,uCi of 35S Trans-label. Cells were then washed with cold 10 mM sodium azide andresuspended in 1 ml of spheroplast buffer (1.4 M sorbitol, 50 mMpotassium phosphate, pH 7.5, 10 mM sodium azide, 50 mM f3-mer-captoethanol, 10 jig of zymolyase/l OD unit of cells). Spheroplastswere formed during a 60-min incubation at 37°C and then lysed in1 ml of lysis buffer (10mM triethanolamine, pH 7.2, 1 mM EDTA, 0.8M sorbitol, and 1 x protease inhibitor cocktail; Ruohola et al., 1988).The unbroken cells were removed during a 3-min spin at 450 x g,and the resulting supematant was heated for 5 min at 100°C in thepresence of SDS (1% final concentration). Samples (100 ,ul) werediluted with 900 ,lI of PBS in 1% Triton X-100 and centrifuged for 15min in an Eppendorf microfuge. The supernatant was removed andincubated overnight on ice with one of the following antibodies: 2 t,lof anti-invertase, 6 ,ul of anti-a-factor, or 2 ,ul of anti-carboxypepti-dase Y. Immune complexes were precipitated onto protein A-Sepha-rose beads (30 ,ul of a 10% solution/,ul of antibody) during a 90-minincubation at 4'C. The beads were washed twice with urea washbuffer (2 M urea, 200 mM NaCl, 100 mM Tris, pH 7.6, 1% TritonX-100), then twice with 1% 3-mercaptoethanol, and the contents ofthe beads were solubilized in 70 ,ul of Laemmli sample buffer.

Polyclonal anti-Bet3p antibody was raised to a Bet3p-MBP fusionprotein that was purified according to manufacturer's directions(New England Biolabs, Beverly, MA). The immunization protocolwas described previously (Louvard et al., 1982) and the serum wasaffinity purified on a column containing cross-linked Bet3p-GSTfusion protein. To precipitate Bet3p, approximately 7 OD units ofcells were resuspended in fresh medium to an OD599 = 3.00 andlabeled at 24°C for 45 min with 0.7 mCi (200 ,Ci/ml) of 35S In Vitrolabel. The protocol used was the same as described above withminor modifications. Spheroplasts were lysed in 0.7 ml of buffer (10mM Tris, pH 8.0, 25 mM EDTA, pH 8.0, and 1% SDS) and thenheated to 100°C for 5 min before they were diluted 1:20 in IP buffer(10 mM Tris, pH 8.0, 150 mM sodium chloride, 0.1 mM ethylenedia-minetetraacetic acid, and 0.5% Tween). The diluted lysates wereincubated with either 5 ,ul of pre-immune serum, 5 ,ul of anti-Bet3pantibody, or 5 .tg/ml of affinity-purified anti-Bet3p polyclonal an-tibody. The immune complexes were precipitated with protein A-Sepharose and the beads were washed twice with IP buffer, twicewith IP buffer containing 2 M urea, and twice with 1% 13-mercap-toethanol.

All samples were subjected to SDS electrophoresis (12.5%). Tocompare strains, the volume of sample loaded onto the gel wasproportional to the incorporation of 5S Trans label in the lysate.Molecular weight standards were as follows: 97.4 kDa, phosphory-lase B; 66 kDa, bovine serum albumin; 45 kDa, ovalbumin; 36 kDa,glyceraldehyde-3-phosphate dehydrogenase; 29 kDa, carbonic an-hydrase; 24 kDa, trypsinogen; 20.1 kDa, trypsin inhibitor; and 14.2kDa, a-lactoalbumin. Gels were stained for protein and fluoro-graphed. Autoradiography was performed using Kodak XAR-5 film(Rochester, NY).

Genetic Techniques and ConstructionsThe BET3 gene was cloned by complementation of the growthdefect of the bet3-1 mutant. The mutant was transformed with ahigh copy plasmid library (Carlson and Botstein, 1982) and thetransformants were selected at 25°C on minimal agar plates lackinguracil. The colonies were then stamped onto YPD plates and incu-bated at 37°C. Complementing plasmids were recovered from yeastby the method of Strathern and Higgins (1991) and subcloned asdescribed below.To amplify plasmid DNA, recombinant plasmids were trans-

formed into DH5a competent cells and isolated as described by

Molecular Biology of the Cell1770

New Yeast Secretory Mutant

Maniatis et al. (1989). Plasmids used in this study (Figure 5) wereconstructed as follows. The original isolate pGR1 contained a14.9-kb insert. Plasmid pGR3 was obtained by subcloning pGR1 asa 5.2-kb Sall-Sall fragment into the Sall site in the polylinker ofpRS426 (Christianson et al., 1992) or the polylinker of pRS306 toyield pGR2. Plasmid pGR4 was constructed in an identical fashionexcept that the Sall-SalI fragment was inserted into the Sall site inthe polylinker of pRS316 (Christianson et al., 1992). To isolate pGR5,the 2.6-kb SaiI-KpnI insert from pGR4 was inserted into the KpnI-Sall sites of pRS316. Plasmid pGR6 was constructed by inserting the1-kb KpnI-PvuII fragment from pGR4 into the KpnI-SmaI sites in thepolylinker region of pRS316. To isolate pGR7, pGR1 was digestedwith PvuII and the vector was religated. Plasmid pGR8 was con-structed by subcloning the 1.55-kb XbaI-XbaI fragment from pGR5into the XbaI site in the polylinker of pRS315.To place the BET3 gene under the control of the regulatable GALl

promoter, polymerase chain reaction (PCR) was used to introducerestriction sites at the amino and carboxy terminus of BET3. ABamHI site was placed two amino acids upstream from the startmethionine of BET3, while an EcoRI site was introduced at theC-terminus of the gene. The PCR product was digested with BamHIand EcoRI before it was subcloned into the BamHI/EcoRI sites in thepolylinker of pNB527. The resulting plasmid (pGR10) was linear-ized with ClaI and integrated into SFNY314 at the LEU2 locus.To disrupt BET3, the KpnI-SaiI fragment shown in Figure 5, was

first subcloned into a bluescript M13 vector in which the HindIII sitewas destroyed. The BET3 gene was then excised from this subclone,by digesting the plasmid with HindIII, and replaced with a HindIII-HindlIl fragment that contains the selectable marker URA3. TheSaII-KpnI fragment was excised, transformed into NY1060, andtetrad analysis was performed.The Bet3p-MBP fusion protein was constructed by introducing a

BamHI site two amino acids upstream from the start methionine ofBET3 using PCR. A BamHI-HindIII fragment was then introducedinto the polylinker site of pPR734 placing BET3 in-frame with the 3'end of the malE gene to yield pGR18. The Bet3-GST fusion proteinwas constructed (pGR20) by introducing a BamHI site two aminoacids from the start methionine of BET3 using PCR. An EcoRI sitewas also introduced after the stop codon of BET3 by PCR. TheBamHI-EcoRI fragment was then subcloned into the BamHI-EcoRIsites of pGEX three times, placing the BET3 gene in-frame with GST.

DNA Sequence Analysis and Homology SearchTo sequence the SaII-KpnI fragment that contains BET3, the 1.6-kbSaiI-PvuII fragment and the 1-kb PvuII-KpnI fragment were sub-cloned into sequencing vectors for single strand DNA sequencing.The 1.6-kb SalI-PvuII fragment was digested from pGR3 and in-serted into the SmaI-SalI sites in the polylinker of pUC118 andpUC119 (Vieira and Messing, 1987). The 1-kb PvuII-KpnI insert wassubcloned into the same vectors using the HincII-KpnI sites of thepolylinker. The DNA sequence was determined by the method ofSanger et al. (1977) using the Sequenase version 2 DNA sequencingkit. The BET3 sequence was compared with other sequences bysearching databases with the TFASTA program.

Enzyme Assays and Other ProceduresWhole cell invertase assays were performed as described by Gold-stein and Lampen (1975); the units of activity are reported as mi-cromoles of glucose released per minute per OD599 of cells. Internalinvertase was assayed as described earlier (Newman and Ferro-Novick, 1987).Samples were prepared for electron microscopy using a modifi-

cation of protocols described previously (Newman and Ferro-Nov-ick, 1987). Briefly, early log phase cells that were grown in YPDmedium were shifted for 2 h at 37°C. The cells were washed in 0.1M cacodylate buffer (pH 6.8), fixed in buffer A (0.1 M cacodylatebuffer, pH 6.8, 3% glutaraldehyde) for 1 h at room temperature, and

then resuspended in fresh buffer A before they were incubatedovernight at 4°C. Fixed cells were washed in 50 mM potassiumphosphate (pH 7.5) and converted to spheroplasts in buffer B (50mM potassium phosphate, pH 7.5, 0.25 mg/ml of Zymolyase lOOT)during a 40-min incubation at 37°C. The spheroplasts were washedtwice with ice cold 0.1 M cacodylate buffer and post-fixed for 1 h at4°C in 0.1 M cacodylate buffer with 2% osmium tetroxide. The pelletwas washed three times with distilled water and then treated witha 2% uranyl acetate solution for 1 h at room temperature. Thesamples were washed twice in water, dehydrated with ethanol,pre-embedded in a 1:1 mixture of acetone with spurr resin, andembedded in spurr resin. Thin sections (60-80 nm) were cut,stained with uranyl acetate and lead citrate, and analyzed with aJEOL-UEM-10 CX II) electron microscope at 80 kV.

RESULTS

Isolation of bet3-1To identify new genes whose defective products arelethal in combination with BETI, a betl-l mutantstrain (SFNY185; Table 1), that harbors BETI on a CENplasmid (pAN102) was mutagenized with ethylmeth-anesulfonate. Because the enrichment of dense cellsleads to the selection of a higher percentage of secre-tory mutants, we used a density enrichment protocolin our screen (Novick et al., 1980; see MATERIAL

Table 1. Yeast strains used in this study

Strain Genotype

ANY113 MATa, ura3-52, his4-619, betl-lSFNY26-3A MATa, ura3-52SFNY26-6A MATa, his4-619SFNY26-12C MATa, ura3-52SFNY185 MATa, ura3-52, his4-619, betl-l, pAN102

(BETI, URA3, CEN)SFNY312 MATa, leu2-3,112, bet3-1SFNY313 MATa, leu2-3,112, bet3-1SFNY314 MATa, ura3-52, leu2-3,112, bet3-1SFNY315 MATa, his4-619, bet3-1SFNY316 MATa, ura3-52, his4-619, bet3-1SFNY317 MATa, ura3-52, his4-619, bet3-1SFNY446 MATa, ura3-52, yptl-3SFNY473 MATa, ura3-52, pGR3 (BET3, URA3, 2,um)NY4 MATa, his4-619, secl-1NY13 MATa, ura3-52NY55 MATa, his4-619, sec9-4NY133 MATa, his4-619, sec2-41NY179 MATa, ura3-52, leu2-3,112NY180 MATa, ura3-52, leu2-3,112sNY404 MATa, his4-619, sec4-8NY411 MATa, his4-619, sec8-9NY413 MATa, ura3-52, secl3-1NY416 MATa, ura3-52, secl6-2NY417 MATa, ura3-52, secl7-1NY419 MATa, ura3-52, secl9-1NY424 MATa, ura3-52, sec21-1NY425 MATa, ura3-52, sec22-3NY432 MATa, ura3-52, secl8-1NY738 MATa, ura3-52, secl2-4NY1060 MATala, GAL/GAL, ura3-52/ura3-52, leu2-3,

112/leu2-3, 112

Vol. 6, December 1995 1771

G. Rossi et al.

AND METHODS). Of the 21,800 colonies analyzed,2,120 were temperature sensitive (ts) for growth at37°C. These mutants were tested for their ability tolose the BETI-containing plasmid (pAN102) on 5-FOAplates. 5-FOA is converted to a toxic compound (5-fluorouracil) by the product of the URA3 gene (oroti-dine-5'-phosphate decarboxylase). In the presence of5-FOA, ura3 (but not URA3) cells will grow. Of the2,120 ts colonies that were tested on these plates, 66failed to grow at 25°C. The remaining strains wereassayed for defects in the secretion of invertase, aglycoprotein that is transported into the yeastperiplasm (Goldstein and Lampen, 1975). This analy-sis revealed that 12 of the 66 ts mutants secreted asignificantly smaller portion of invertase than wildtype (' 50%) at 37°C. Additionally, all the mutantsaccumulated active invertase.Linkage studies and plasmid complementation anal-

ysis were performed with the remaining mutants todetermine whether any were defective in previouslyidentified genes whose products have been implicatedin secretion. These studies indicated that ts 44 wasallelic with sec22, while five other mutants (ts 14, 20a,40, 60, and 75) harbored mutations in three known secgenes (sec3, sec4, and secl7). These strains, as well asthe remaining mutants, were mated to wild-type cells.Further genetic analysis was performed by displacingplasmid pAN102 and sporulating the diploid cells.The betl-1 mutation was then separated from thenewly identified ts mutation during tetrad analysis.Only one mutant, ts29, was found to be defective in agene that is lethal in combination with betl-1. In con-trast, the sec3, sec4, and secl7 alleles that were identi-fied by this selection were not lethal with betl, sug-gesting that their isolation was fortuitous. Theremaining five mutants that accumulated invertasedid not harbor mutations that were lethal in combina-tion with betl-1.

To determine whether the defect in ts29 was dueto one mutation, this strain was crossed to wildtype, sporulated, and subjected to tetrad analysis.Of the 50 tetrads examined, the growth defect seg-regated 2:2 in all cases, indicating that it was due toa single ts mutation. Furthermore, the secretion de-fect co-segregated with the growth defect in three ofthe tetrads examined. To demonstrate that ts29 islethal in combination with betl-1, we crossed one ofthe ts colonies to betl-l and performed tetrad anal-ysis. Twelve tetrads were examined. In each case,the double mutants were inviable (see Table 3),demonstrating that ts29 is defective in a new genewhose product interacts genetically with BET1. Wecalled this new mutant bet3-1. Subsequent analysisof bet3 was performed with a strain that was back-crossed twice to wild type.

The Secretion of Invertase Is Rapidly Blocked inbet3-1 upon a Shift to 37°COur analysis of invertase indicated that this markerprotein failed to be transported to the periplasm at37°C (Table 2), although transit appeared to be nor-mal (our unpublished results) at the permissivetemperature (24°C). If BET3 plays a direct role insecretion, then cells harboring a mutation in thisgene should rapidly block the export of invertaseafter a shift to the restrictive temperature. To assessthe speed at which the bet3 mutant phenotype isexpressed at 37°C, cells were first incubated for 35min at 25°C in low glucose-containing medium toderepress the synthesis of invertase. Mutant cellswere then shifted to 37°C and internal and externallevels of the enzyme were measured at various time-points. We found that active invertase accumulatedwithin the bet3-1 mutant 5 min after it was incu-bated at its restrictive temperature (Figure 1A). Fur-

Table 2. Reversibility of accumulated invertase (U/OD599)

40 min (37°C)

40 min (37°C) 3h (250C) 25°C

Strain Description External Internal External Release

SFNY 26-6A Wild-type 0.39 0.08 ... ...

NY 432 sec 18-1 0.02 1.1 0.92 82%ANY 113 bet 1-1 0.1 0.53 0.49 74%SFNY 312 bet 3-1 0.12 0.9 0.39 30%

Cells were grown overnight in YPD medium at 250C to an early exponential phase and then incubated at 37°C for 10 min. At the end of thisincubation, the cells were resuspended in 2.0 ml of prewarmed YP medium containing 0.1% glucose (OD599= 1.0) and incubated for 40 minlonger at the same temperature. Aliquots (1.0 ml) were pelleted, resuspended in prewarmed YPD medium with 0.1 mg/ml of cycloheximide,and incubated at 250C for 3 h; the remaining cells were placed on ice. Subsequent to all incubations, the cells were washed with cold 10 mMsodium azide, and stored on ice until they were assayed for internal and external levels of invertase. The percent release at 25°C wascalculated by the following equation: [external (37°C-25°C)-40 min external invertase (370C)]/40 min internal invertase (37°C).

Molecular Biology of the Cell1772

New Yeast Secretory Mutant

thermore, only a small amount of invthe initial block (Figure 1B). Thus,reported for betl-l (Newman et al., 1'mutant phenotype is rapidly displayEto the restrictive temperature.

A.--

in

C:V

4)

B

tr_

O-*

ID

Q

4)

4)-

0 10 20 30 40Time (min)

0.14

0.12

0.10

0.08

0.06

0.04Shift to0.1% glucoseI

0.024/

00 10 20 30 40

Time (nmmn)

Figure 1. The bet3-1 mutant quickly displays asecretion upon a shift to the restrictive tempegrown overnight at 25°C in YPD medium tophase. Samples were resuspended in YP mediu:glucose at 25°C. At the indicated time points, 1.removed, washed, resuspended in cold 10 mMstored on ice. After 35 min at 25°C, the remainileted, resuspended in prewarmed YP medium cccose, and incubated at 37°C. Aliquots were againdicated time points, washed, and resuspendedazide. Internal (A) and external (B) levels of inveas described in MATERIALS AND METHODS.

'ertase escapedas previously)90), the bet3-1

____ 1

The bet3-1 Mutant Pleiotropically Blocks theTransit of Proteins from the ER to the GolgiComplex

ed upon a snhtt A prediction of the synthetic lethal screen that weemployed is that bet3 should be defective in ER toGolgi transport. Because the precursor form of a pro-tein that accumulates intracellularly is indicative ofthe stage at which transport is blocked (Novick et al.,1981; Newman and Ferro-Novick, 1987), we radiola-beled yeast cells during the derepression of invertaseand examined the form that is retained within the cell(Carlson and Botstein, 1982). The secreted form of

370C invertase is initially synthesized as a 62-kDa protein.In the lumen of the ER, the signal sequence is cleavedand 9-10 N-linked oligosaccharide units are added toinvertase to yield an -80-kDa species. As has beenshown for betl-l (Newman and Ferro-Novick, 1987)

- 250C and other ER-accumulating mutants, such as secl8-1(Novick et al., 1981; see control in Figure 2A, lane 3),the invertase that failed to be secreted in the bet3-1

l__ mutant at 37°C accumulated as an 80-kDa species50 60 70 (Figure 2A, compare lane 2 with wild type in lane 1).

To determine the extent to which the 80-kDa form ofinvertase is reversibly secreted, bet3-1 mutant cellswere first shifted to 37°C in low glucose-containingmedium to accumulate precursor intracellularly (Ta-ble 2). Subsequent to this incubation, the cells wereshifted to 24°C for 3 h in the presence of cyclohexi-mide, a drug that inhibits further protein synthesis. Tocontrol for the leakage of accumulated enzyme, a par-

250C allel experiment was performed at 370C (our unpub-lished results). The data in Table 2 indicates that thebet3-1 mutant secreted invertase upon a shift to thepermissive temperature. The secretion of invertasewas greater at 240C than 37°C, indicating that thedefect in transport was partially reversible.Transport of the yeast pheromone a-factor and the

vacuolar protease carboxypeptidase Y (CPY) were also370C examined in bet3-1. a-Factor is synthesized as a 19-

kDa precursor that is converted to a 26-kDa species inthe lumen of the ER. The 26-kDa ER form of a-factoraccumulates in the betl-1, secl8-1 (Newman andFerro-Novick, 1987; Figure 2B, lanes 3 and 4), andbet3-1 mutants at 370C (Figure 2B, compare lanes 2, 3,

50 60 70 and 4 with wild type in lane 1). A partially glycosy-lated form of a-factor, which contains two instead of

block in invertase three N-linked oligosaccharide units, was also ob-rature. Cells were served in these mutants. Furthermore, like betl-l (Fig-early exponential ure 2C, lane 3) and secl8-1 (Figure 2C, lane 4), bet3-1

0m contamiingqOs (Figure 2C, lane 2) accumulates the pl form (67 kDa)sodium azide, and of CPY. pl CPY is further modified in the Golgi ap-ing cells were pel- paratus to p2 CPY (69 kDa) before it is processed into)ntaining 0.1% glu- the mature form (61 kDa; Figure 2C, lane 1) in the

Iin 10 mM sodium vacuole (reviewed by Klionsky et al., 1990). Thus, atrtase were assayed 37°C bet3-1 displays a pleiotropic block in membrane

traffic.

Vol. 6, December 1995 1773

G. Rossi et al.

B

-- 80OkD

1 2 3

, -4--- 26kD

1 2 3 4

C

a~IpCPY

1 2 3 4

Figure 2. The bet3-1 mutant fails to transport invertase, a-factor,and CPY from the ER to the Golgi complex. Mutant (lanes 2, 3, and4) and wild-type (lane 1) cells were labeled with 35S Trans label at37°C. Radiolabeled cells were converted to spheroplasts and lysedas described in MATERIALS AND METHODS. Invertase (A), a-fac-tor (B), and CPY (C) were immunoprecipitated from each sampleand analyzed on a 12.5% SDS polyacrylamide gel. ER forms ofinvertase (-80 kDa), a-factor (26 kDa), and CPY (pl) were found toaccumulate in bet3-1 (lane 2), betl-l (lane 3 in panels B and C), andsecl8-1 (lane 3 in panel A and lane 4 in panels B and C) mutant cells,but not wild type (lane 1).

The bet3-1 Mutant Accumulates ER and Vesicles atIts Restrictive Growth TemperatureThe betl-l mutant accumulates ER membrane at itsrestrictive growth temperature (Newman and Ferro-Novick, 1987). To morphologically examine bet3-1 at37°C, thin sections of this mutant were compared withwild type. In wild-type cells, tubules of ER are occa-sionally found at the periphery of the cell or juxta-posed to the nuclear envelope. In contrast, whenbet3-1 was incubated at 37°C for 2 h, the ER membranewas elaborated (Figure 3, compare A and B) and somesmall vesicles (average size -60 nm) were found toaccumulate in the cytoplasm (Figure 3B). The lumen ofthe ER was dilated and sometimes the nucleus ap-peared to be multi-lobed. The betl-l mutant also ac-cumulates small vesicles at 37°C. These vesicles arecomparable in size to the vesicles that accumulate insecl7, secl8, sec22 (Novick et al., 1980), and Boslp-depleted cells (Shim et al., 1991). Thus, the bet3-1mutant is phenotypically similar to betl-l at its restric-

tive growth temperature (Newman and Ferro-Novick,1987). In addition to small vesicles, bet3-1 also accu-mulated a number of large vesicles (average size -110nm). Vesicles of this size have been seen before insecretory mutants that disrupt post-Golgi secretion(Novick et al., 1980). Both small and large vesiclesaccumulate in bet2-1 and secl9-1, two mutants thatfunction at more than one stage of the yeast secretorypathway (Newman and Ferro-Novick, 1987; Rossi etal., 1991; Garrett et al., 1994).

Like BET1, BET3 Interacts Genetically with BOS1,SEC21, SEC22, and YPT1Previous studies have shown that BETI interacts ge-

netically with BOS1, SEC21, SEC22, and YPT1 (New-man et al., 1990, 1992; Dascher et al., 1991). To deter-mine whether BET3 also interacts with members ofthis group of genes, we transformed the bet3-1 mutantwith either the BOSi, SEC21, SEC22, YPT1, or BETIgene on a multicopy plasmid and compared thegrowth of the transformed strain to the parent at var-ious temperatures. These studies revealed that al-though bet3-1 failed to grow at 30°C, the overproduc-tion of either BOSi, SEC22, YPT1, or BETI facilitatedgrowth at this temperature (Figure 4). In contrast, itwas difficult to transform the bet3-1 mutant with theSEC21 gene on a multicopy plasmid. However, whentransformants were obtained, they grew slower thanthe parent strain (our unpublished results). Thus, theoverproduction of either of these genes enhances orinhibits the growth of the bet3-1 mutant, suggestingthat they interact with BET3. In support of these find-ings, we have also found that the following haploiddouble mutant combinations were inviable at 25°C:bet3-1 and betl-1; bet3-1 and sec21-1; bet3-1 andsec22-3; and bet3-1 and yptl-3 (see Table 3). A tsmutant in bosl was not available for this analysis.These synthetic lethal interactions are specific as mu-tations in two genes (secl3-1 and secl6-2), whoseproducts are required for vesicle budding, are notlethal in combination with betl-1.

Because thin section electron microscopy has re-vealed that the bet3-1 mutant accumulates both smalland large vesicles (Figure 3), the BET3 gene productmay function at more than one stage of the yeastsecretory pathway. To explore this possibility further,we crossed the mutant to three ER-accumulating mu-

tants, seci7-1, seci8-1, and secl9-1 that have beenproposed to function at multiple stages of the path-way, as well as several post-Golgi mutants (sec2-41,sec4-8, sec8-9, and sec9-4). This analysis has shownthat haploid double mutants containing bet3-1 andsecl7-1, bet3-1 and secl8-1, bet3-1 and sec2-41, andbet3-1 and sec4-8 were either sick or inviable at 25°C.Thus, these genetic studies support the morphological

Molecular Biology of the Cell

A

1774

New Yeast Secretory Mutant

studies that suggest Bet3p may function in more thanone transport event.

Cloning the BET3 GeneTo clone a wild-type copy of the BET3 gene, the bet3-1mutant (SFNY316) was transformed with a library ofyeast genomic inserts (Carlson and Botstein, 1982)cloned into a high copy vector, YEp24. Of the 30,000colonies screened, seven grew at 37°C. Plasmids re-covered from six of these strains fully complementedthe growth defect of bet3-1 at 37°C. The analysis ofthese six plasmids defined a common 5.2-kb Sall-Sallfragment (pGR4 in Figure 5). To determine whetherthis insert contained the BET3 structural gene, an in-tegration experiment was performed. The 5.2-kb Sall-Sall fragment was subcloned into an integrating vector(pGR2) and digested with KpnI to direct integrationinto the genome of the bet3-1 mutant (SFNY316). Thisevent placed URA3 adjacent to the cloned gene. Theresulting strain was backcrossed to wild type (NY179)and the diploid was sporulated. Of the 16 tetradsexamined, the Ts' colonies were always Ura+, dem-onstrating that the cloned locus was tightly linked toBET3.Subcloning studies (see Figure 5) revealed that BET3

was contained within a smaller 2.6-kb SalI-KpnI frag-ment (pGR5). When this region was further subclonedto a 1-kb PvuIL-KpnI fragment (pGR6), there was a lossof complementing activity. Furthermore, when the3.6-kb region to the right of the PvuIl site was deletedfrom pGR4 (see Figure 5), activity was also abolished.Thus, the 1.6-kb SalI-PvuII fragment (pGR7) was notsufficient to complement the growth defect of thebet3-1 mutant. Based on these subcloning studies, weconclude that the PvuII site is contained within theBET3 gene. In support of this proposal, we haveshown that the 1.55-kb XbaI-XbaI fragment (pGR8 in

Figure 5), which contains this PvuII site, is sufficient tocompletely restore growth to bet3-1 at 37°C.

Nucleotide Sequence of the BET3 GeneAs a first step toward determining the function ofBET3, we cloned and sequenced the gene that encodesthis protein. Because the PvuII site described above iscontained within BET3, we subcloned the SalI-PvuII(1.6 kb) and PvuII-KpnI (1 kb) fragments into thepUC118 and pUC119 vectors and sequenced to the leftand right of the PvuII site. This analysis revealed thatone open reading frame spans the PvuII site. Thesequence of the -0.6-kb HindIll-HindIll fragment,which contains this open reading frame, is locatedon chromosome XI (accession number Z28292-4;Dujon et al., 1994). A search of the data bank re-vealed that BET3 is 36.4% identical and 60% similarto a 20-kDa (p20) nematode protein (accession num-ber P34605) of unknown function (Figure 6), indi-cating that homologues of Bet3p can be found in avariety of organisms.To confirm that the identified open reading frame

encodes BET3, its ability to complement the growthdefect of the bet3-1 mutant was tested. This was doneby cloning the open reading frame behind the regu-latable GALI promoter. Transcription from this pro-moter is repressed in the presence of glucose, but notgalactose. The resulting plasmid (pGR10 in Figure 5)was integrated at the LEU2 locus of SFNY314 (seeTable 1) and two Leu+ transformants were tested fortheir ability to grow on plates containing galactose orglucose. These transformants only grew on YP galac-tose plates at 37°C, confirming that the HindIII-Hin-dIll fragment that we sequenced encodes the BET3gene.

Table 3. Tetrad analysis of crosses with bet3-1

Cross No. of crosses

bet3-1 Four viable Three viable Two viablecrossedwith: 4-o:+a 3-:1+ 3-:O+ 2-:1+ 1-:2+ 1-:1+ 0-:2+

betl-1 1 11sec2-41 2 9 1sec4-8b 1 9 1secl7-1b 11sec18-1b 1 10 1sec21-1 8 1 1 8 1 1 3sec22-3 3 4 2 3yptl-3 5 1 10 2 6

a Designates the number of spores that grew (+) or did not grow (-) at 37°C.b Double mutants were either sick or lethal.

Vol. 6, December 1995 1775

G. Rossi et al.

AerX

t7

51-~~ ~ ~ ~~e e....

Z'4- '

.X.s, s~tER~~~~~~~~~*s ..

44

Bsvse.a_; <; xr,..,jst,t.s

i.gg<'i '' '.

<s :r,: , ,.' <4.^.'*iAi* . .. + r t>.* .. . e - . s - :I Ss.. : .: ffle, , ,', , W-';;: b 'BU

::

., ,.w; .:. . e : . .. §.... . , .+ is:s .. :,.' .. i, k JY;

iS 11E W-Fi,^:>.: X 1!>.r

-,. .., gr:,-; tt W:. . ., . > > I>wZi.*: '' . ""* :' . t ' ,' ' . <:., ' '. -;,.

:Y: ,.;

.:

>,: 4ee;s.. .. Si t.E.t...

.E;D,U,;

:w

__S

.,._k

1 .v

o

, 'r@, '. . . . . .

BET3 Is Essential for FunctionTo determine whether BET3 encodes a protein that isessential for the vegetative growth of yeast cells, wedisrupted the BET3 open reading frame in a diploidstrain (NY1060) using URA3 as a selectable marker. Aseries of constructions were performed before disrupt-ing the genomic copy of BET3. First, we subcloned the2.6-kb SaII-KpnI fragment shown in Figure 5 into a

CM- E Figure 3. EM analysis of bet3-1 mutantcells. Samples were prepared for electronmicroscopy as described in MATERIALSAND METHODS. Thin-section analysis ofwild-type (A) and bet3 mutant cells (B).ER, endoplasmic reticulum; N, nucleus; V,vesicles; SV, small vesicles; Bar, -0.5 Am.

bluescript vector that lacked a HindIII site (see MA-TERIALS AND METHODS). Subsequently, this vectorwas digested with HindIII, removing the 0.6-kb Hin-dIII-HindIII fragment that contains BET3. The URA3gene was then inserted in its place to yield plasmidpGR 9. This plasmid was digested with KpnI and Sall,integrated into a diploid strain (NY1060), sporulated,and tetrad analysis was performed. Of the 24 tetrads

Molecular Biology of the Cell1776

New Yeast Secretory Mutant

d

e

f

aFigure 4. The overexpression of BETI, BOSI, SEC22, or YPT1suppresses the growth defect of the bet3-1 mutant at 30'C. Thebet3-1 mutant (SFNY316) was transformed with either (a) pRB307(vector alone; URA3, 2 ,um); (b) pAN109 (BOS1, URA3, 2 ,um); (c)pFN100 (BET1, URA3, 2 ,im); (d) pJG103 (SEC22, URA3, 2 t,m); or(e) pNB167 (YPT1, URA3, 2 jim). The growth of the transformantswere compared with wild type (f) on a YPD plate at 30'C.

examined, all displayed 2:2 segregation for viability.In addition the viable spores were Ura-, indicatingthat BET3 is required for the viability of yeast cells. Todemonstrate that the inviable spores failed to growbecause they lacked Bet3p, we introduced pGR10

(GAL1::BET3) into cells disrupted for BET3. Growthwas observed on YP galactose plates, but not on YPDplates. Thus, the growth defect was only rescuedwhen BET3 was expressed.

BET3 Encodes a 22-kDa ProteinThe nucleotide sequence of BET3 predicts a hydro-philic protein of 22-kDa with a net negative charge of6.9 at neutral pH. To confirm the size of Bet3p, poly-clonal antibody was raised to a Bet3p-MBP fusionprotein. Lysates prepared from cells that were radio-labeled by the protocol described in MATERIALSAND METHODS were precipitated with pre-immuneserum (Figure 7, lane 1 and 2) or a-Bet3p (Figure 7,lane 3). This analysis revealed that a-Bet3p recognizesa 22-kDa band, which was also overproduced in astrain that overproduces Bet3p (Figure 7, lanes 4 and6). Affinity-purified a-Bet3p cross-reacted with thesame protein species (Figure 7, lane 5). A c-myc-tagged version of Bet3p, of approximately the samemolecular weight, was also recognized by this anti-body (our unpublished results).

The SNARE Complex Does Not Form in bet3-1Mutant CellsGenetic and morphological studies imply that Bet3pmay play a role in the late stages of ER to Golgitransport. With antibody in hand, we directly testedthe possibility that Bet3p is a component of theSNARE complex. When vesicle fusion is blocked at37°C in secl8 mutant cells, the SNARE complex accu-mulates. This complex has been immunoisolated fromdetergent extracts of secl8 with affinity-purified anti-Boslp antibody (Sogaard et al., 1994). It contains thet-SNARE, which is proposed to be Sed5p, the

Complementation

A pGR4 S X H P H X KI I I I I

B pGR52.6kb

S.bet3-1 A bet3

+ n.d.

+ n.d.

C pGR6

D pGR7

Figure 5. Complementingactivity of cloned inserts.Plasmids were prepared asdescribed in MATERIALSAND METHODS. Only thecloned inserts are shown. S,Sall; X, XbaI; H, HindIII; P,PvuII; K, and KpnI. bp, basepairs.

E pGR8

F PGRlo

- n.d.lkb

- n.d.1.6kb

1.55kb

Gal I0.58kb

500bp

Vol. 6, December 1995

+ +

+ +

1777

G. Rossi et al.

Bet3p

p20

Bet3p

p20

Bet3p

p20

Bet3p

p20

13 MGEEIWKNKTE..KINTELFTLTYGSIVAQLCQDYERDFNKVNDHLYSMG 60

1 MSKAIKQNLADSKKMSAELFCLTYGAMVMTELDYE.DPKDVTIQLDKMG 49

61 YNIGCRLIEDFLARTA.LPRCENLVKTSEVLSKCAFKIFLNITPNITNWS 109

50 FNMGTRLADDFLAKNANVPRCVDTRQIADVLCRNAIPCYLGISATASSWT 99

110 HNKDTFSLILDENPLADFVELPMDAMK.SLWYSNILCGVLKGSLEMVQLD 158

100 SGDREFTITLEANPLTELVQVPAHLVSAGLSYSQLIAGAIRGALEAVHFK 149

159 CDVWFVSDILRGDSQTEIKVKLNRILKDEIPIGED 193

150 V...YASAT.DTGANTEIRIRFDQVLKDSLPAGED 180

v-SNARE, and several unidentified proteins that aresimilar in molecular weight to Bet3p. In performingour studies, we prepared a detergent extract of seci8mutant cells and precipitated it with a-Boslp. As acontrol, we also examined wild-type extracts that donot accumulate this docking/fusion complex. The pre-cipitate was then blotted with antibodies to Sec22pand Sed5p, which are components of the SNARE com-plex, as well as Bet3p. As expected, a-Boslp precipi-tated Sec22p and Sed5p from the secl8 mutant extract,but not wild type (Figure 8A, lanes 1 and 2). Bet3p wasnot found in this precipitate (Figure 8A, lanes 1, 2, and3), however, it was recovered in the supernatant frac-tion (our unpublished results).Although the findings described above indicate that

Bet3p is not a component of the SNARE complex,active Bet3p may be required to form this complex. Totest this possibility, we precipitated an extract ofbet3-1 mutant cells with a-Boslp and blotted the pre-cipitate with a-Sec22p and a-Sed5p. As shown in Fig-ure 8B, Boslp co-precipitated with Sec22p and Sed5p

kD29-24-

420-

Figure 6. A comparison of Bet3p with the C.elegans protein p20. The amino acid sequence ofBet3p is compared with p20 using the Bestfitprogram (GCG software package). Identity is in-dicated by a line between the sequences andconserved changes are indicated by two dots(two corresponding bases in a codon) or one dot(one corresponding base in a codon). Gaps areshown as dots within the sequence. Bet3p andp20 are 36.4% identical at the amino acid level.

from the secl8-1 mutant extract (Figure 8B, lane 2),but not from wild type (Figure 8B, lane 1) or the bet3-1extract (Figure 8B, lane 3). Thus, the SNARE complexfails to form in bet3-1 mutant cells. Furthermore, theBoslp/Sec22p complex, which accumulates in secre-tory mutants that are blocked in vesicle targeting andfusion (Lian et al., 1994), could not be detected in the

A

Sec22p --

Bet3p1 2

Sed5p- w1 2 3

1 2

B

Sec22p- _

1 2 3

- Sed5p

1 2 3

1 2 3 4 56

Figure 7. BET3 encodes a 22-kDa protein. Wild-type cells (NY13)or cells overexpressing the BET3 gene (SFNY473) were radiolabeled,spheroplasted, and lysed as described in MATERIALS AND METH-ODS. The lysates were incubated with either pre-immune serum(lanes 1 and 2), anti-Bet3p antibody (lanes 3 and 4), or affinity-purified anti-Bet3p antibody (lanes 5 and 6) and the precipitateswere analyzed on a 12.5% SDS gel. A 22-kDa protein was detectedby the anti-Bet3p antibody, but not by the pre-immune serum. The22-kDa band was also overproduced in a strain that overproducesBET3 (lanes 4 and 6).

Figure 8. Bet3p is not a component of the yeast SNARE complex.Wild type (lane 2 in panel A and lane 1 in panel B), secl8-1 (lane 1in panel A and lane 2 in panel B), and bet3-1 (lane 3 in panel B) cellswere grown in YPD medium to early log phase. Cells were con-verted to spheroplasts at 25'C using protocols described previously(Ruohola et al., 1988) and the spheroplasts were regenerated for 1 hat 37'C. Regenerated spheroplasts were lysed, and processed asdescribed elsewhere (Lian et al., 1994). The clarified lysates wereprecipitated with 40 jig of affinity-purified a-Boslp and the solubi-lized precipitates were blotted with a-Sec22p and a-Sed5p. A lysatethat overproduces Bet3p marks the location of this protein in panelA (lane 3).

Molecular Biology of the Cell1778

New Yeast Secretory Mutant

bet3-1 mutant extract (Figure 8B, compare lanes 1 and3 with lane 2). The formation of the Boslp/Sec22pcomplex correlates with the ability of Boslp to pairwith Sed5p (Lian et al. 1994).

DISCUSSION

Here we report the use of a synthetic lethal screen toidentify an additional gene whose defective product islethal in combination with betl-1. This analysis hasresulted in the isolation of bet3-1, a new ts yeast se-cretory mutant that fails to transport proteins from theER to the Golgi complex at 37°C. The phenotypicconsequences of mutations in bet3-1 are similar tothose that have been observed previously for betl-1.Furthermore, like BETI, BET3 displays genetic inter-actions with BOSi, SEC21, SEC22, and YPT1. Thus,Bet3p may act in conjunction with Betlp to mediatethe same step in ER to Golgi transport.Although BET3 interacts genetically with a group of

genes whose products are components of the SNAREcomplex, the Bet3 protein is not physically associatedwith this complex. Furthermore, the Boslp/Sec22pcomplex, as well as the SNARE complex, does notaccumulate in bet3-1 mutant cells. These findings sup-port the hypothesis that Bet3p acts upstream of thisdocking/fusion complex. A clue to the role of Bet3phas come from studies with Betlp and Boslp. Boslpspecifically functions on vesicles where it forms acomplex with Sec22p, a protein that modulates itsactivity (Lian and Ferro-Novick, 1993; Lian et al.,1994). These interactions are catalyzed by the ras-likeGTP-binding protein Yptlp. Betlp may interact withBoslp on the ER before Boslp associates with Sec22pon vesicles. Thus, it is possible that Yptlp stimulatesthe formation of the Boslp/Sec22p complex throughBetlp, and Bet3p may act in concert with it. An anal-ysis of Bet3p in an in vitro reaction that reconstitutestransport from the ER to the Golgi complex (Ruoholaet al., 1988; Groesch et al., 1990) will enable us toaddress this possibility.Two findings suggest that Bet3p may function at

more than one stage of the secretory pathway. First, amorphological analysis of bet3-1 has revealed thatsmall (mean size -60 nm) and larger (average size-110 nm) vesicles accumulate in this mutant at 37°C.The larger vesicles are similar to those seen previouslyin mutants that are blocked in post-Golgi transport(Novick et al., 1980). Second, genetic studies have re-vealed that bet3-1 is lethal in combination withsec2-41 or sec4-8, two mutants that are defective inthe fusion of post-Golgi transport vesicles with theplasma membrane (Novick et al., 1980). A predictionof these observations is that BET3 may encode a sol-uble protein that functions in both ER to Golgi andpost-Golgi transport. In support of this hypothesis isthe finding that BET3 encodes a highly conserved

22-kDa hydrophilic protein. Further characterizationof Bet3p will allow us to determine whether this pro-tein and its homologues function at more than onestage of the secretory pathway in yeast and highercells.The findings described here have shown that syn-

thetic lethal screens can be used to isolate secretorymutants that are phenotypically related to each other.The advantage of such a screen is that it facilitates theidentification of physically interacting components, orgene products that function on the same or parallelpathways. Now that we have demonstrated that thisapproach will lead to the isolation of new bet mutants,we are currently using similar screens to find addi-tional genes whose products functionally interact withBoslp and Sec22p.

ACKNOWLEDGMENTSWe thank H. Pelham, D. Botstein, M. Yamasaki, R. Schekman, andH.D. Schmitt for strains and plasmids, P. Brennwald for criticaladvice throughout the course of this work, Y. Jiang and P. Lyons forassistance in tetrad analysis, F. Finger for advice on the syntheticlethal screen, and Anne Marie Quinn for DNA sequence analysis.We also thank M. Garrett for comments on the manuscript andHenry Tan for photography. Guendalina Rossi was the recipient ofan Argall L. and Anna G. Hull Cancer Research Award and aPatrick and Catherine Weldon Donaghue Medical Research Foun-dation fellowship. Francois Palluault was the recipient of a JamesHudson Brown-Alexander B. Coxe postdoctoral fellowship. Thiswork was supported by grants awarded to S.F.-N. from the NationalInstitutes of Health (1 RO1 GM-45431 and CA-46128).

REFERENCES

Carlson, M., and Botstein, D. (1982). Two differentially regulatedmRNAs with different 5' ends encode secreted and intracellularforms of yeast invertase. Cell 28, 145-154.

Christianson, T.W., Sikorski, R.S., Dante, M., Shero, J.H., and Hieter,P. (1992). Multifunctional yeast high copy number shuttle vectors.Gene 110, 119-122.

Clary, D.O., Griff, I.C., and Rothman, J.E. (1990). SNAPs, a family ofNSF attachment proteins involved in intracellular membrane fusionin animals and yeast. Cell 61, 709-721.

Dascher, C., Ossig, R., Gallwitz, D., and Schmitt, H.D. (1991). Iden-tification and structure of four yeast genes (SLY) that are able tosuppress the functional loss of YPT1, a member of the RAS super-family. Mol. Cell. Biol. 11, 872-885.

Dujon et al. (1994). Complete DNA sequence of yeast chromosomeXI. Nature 369, 371-378.

Ferro-Novick, S., and Jahn, R. (1994). Vesicle fusion from yeast toman. Nature 370, 191-193.

Ferro-Novick, S., and Novick, P. (1993). The role of GTP-bindingproteins along the exocytic pathway. Annu. Rev. Cell Biol. 9, 575-599.

Finger, F.P., and Novick, P.J. (1992). Isolation of new secretorymutants that are synthetically lethal with sec4-8. Mol. Biol. Cell 3,309a.

Garrett, M.D., Zahner, J.E., Cheney, C.M., and Novick, P.J. (1994.)GDI1 encodes a GDP dissociation inhibitor that plays an essentialrole in the yeast secretory pathway. EMBO J. 13, 1718-1728.

Vol. 6, December 1995 1779

G. Rossi et al.

Goldstein, A., and Lampen, J.O. (1975). ,B-D-Fructofuranoside fruto-hydrolase from yeast. Methods Enzymol. 42, 504-511.Graham, T.R., and Emr, S.D. (1991). Compartmental organization ofGolgi-specific protein modification and vacuolar protein sortingevents defined in a yeast secl8 (NSF) mutant. J. Cell Biol. 114,207-218.Griff, I.C., Schekman, R., Rothman, J.E., and Kaiser, C.A. (1992). Theyeast SEC17 gene product is functionally equivalent to mammaliana-SNAP. J. Biol. Chem. 267, 12106-12115.Groesch, M., Ruohola, H., Bacon, R., Rossi, G., and Ferro-Novick, S.(1990). Isolation of a functional vesicular intermediate that mediatesER to Golgi transport in yeast. J. Cell Biol. 111, 45-53.Hosobuchi, M., Kreis, T., and Schekman, R. (1992). SEC21 is a generequired for ER to Golgi transport that encodes a subunit of yeastcoatomer. Nature 360, 603-605.Klionsky, D.J., Herman, P.K., and Emr, S.D. (1990). The fungalvacuole: composition, function, and biogenesis. Microbiol. Rev. 54,266-292.Lian, J.P., and Ferro-Novick, S. (1993). Boslp, an integral membraneprotein of the ER to Golgi transport vesicles, is required for theirfusion competence. Cell 73, 735-745.Lian, J.P., Stone, S., Jiang, Y., Lyons, P., and Ferro-Novick, S. (1994).Yptlp implicated in v-SNARE activation. Nature 372, 698-701.Louvard, D., Reggio, H., and Warren, G. (1982). Antibodies to theGolgi complex and rough endoplasmic reticulum. J. Cell Biol. 92,92-107.Malhotra, V., Orci, L., Glick, B.S., Block, M.R., and Rothman, J.E.(1988). Role of an N-ethylmaleimide-sensitive transport componentin promoting fusion of transport vesicles with cisternae of the Golgistack. Cell 54, 221-227.

Maniatis, T., Fritsch, E.F., and Sambrook, J. (1989). Plasmid vectors:extraction and purification of plasmid DNA. In: Molecular Cloning:A Laboratory Manual, vol.I, 1.25-1.28.Newman, A., and Ferro-Novick, S. (1987). Characterization of newmutants in the early part of the yeast secretory pathway isolated bya [3H]mannonse suicide selection. J. Cell Biol. 105, 1587-1594.

Newman, A.P., Groesch, M., and Ferro-Novick, S. (1992). Boslp, amembrane protein required for ER to Golgi transport in yeast,co-purifies with carrier vesicles and with Betlp and the ER mem-brane. EMBO J. 11, 3609-3617.

Newman, A.P., Shim, J., and Ferro-Novick, S. (1990). BET1, BOS1and SEC22 are members of a group of interacting genes required fortransport from the ER to the Golgi complex. Mol. Cell. Biol. 10,3405-3414..Novick, P., Ferro, S., and Schekman, R. (1981). Order of events in theyeast secretory pathway. Cell 25, 461-469.Novick, P., Field, C., and Schekman, R. (1980). Identification of 23complementation groups required for post-translational events inthe yeast secretory pathway. Cell 21, 205-215.Rossi, G., Jiang, Y., Newman, A.P., and Ferro-Novick, S. (1991).Dependence of Yptl and Sec4 membrane attachment on Bet2. Na-ture 351, 158-161.Ruohola, H., Kabcenell, A.K., and Ferro-Novick, S. (1988). Reconsti-tution of protein transport from the endoplasmic reticulum to theGolgi complex in yeast: the acceptor compartment is defective in thesec23 mutant. J. Cell Biol. 107, 1465-1476.Sanger, F., Nicklen, S., and Coulson, A.R. (1977). DNA sequencingwith chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74,5463-5467.Shim, J., Newman, A.P., and Ferro-Novick, S. (1991). The BOSi geneencodes an essential 27-kD putative membrane protein that is re-quired for vesicular transport from the ER to the Golgi complex inyeast. J. Cell Biol. 113, 55-64.Sogaard, M., Tani, K., Ye, R.R., Geromanos, S., Tempst, P., Kirch-hausen, T., Rothman, J.E., and Sollner, T. (1994). A rab protein isrequired for the assembly of SNARE complexes in the docking oftransport vesicles. Cell 78, 937-948.S6llner, T., Whiteheart, S.W., Brunner, M., Erdjument-Bromage, H.,Geromanos, S., Tempst, P., and Rothman, J.E. (1993). SNAP recep-tors implicated in vesicle targeting and fusion. Nature 362, 318-324.

Strathern, J.N., and Higgins, D.R. (1991). Recovery of plasmids fromyeast into Escherichia coli: shuttle vectors. In: Guide to Yeast Geneticsand Molecular Biology, vol. 194, 319-329.Vieira, J., and Messing, J. (1987). Production of single-strandedplasmid DNA. Methods Enzymol. 153, 3-11.Warren, G. (1993). Bridging the gap. Nature 362, 297-298.Wilson, D.W., Wilcox, C.A., Flynn, G.C., Chen, E., Kuang, W.-J.,Henzel, W.J., Block, M.R., Ullrich, A., and Rothman, J.E. (1989). Afusion protein required for vesicle-mediated transport in both mam-malian cells and yeast. Nature 339, 355-359.

Molecular Biology of the Cell1780